Research.
Research in machine perception, language, and efficient learning. From a better understanding to a working system.
Google Scholar ↗Download BibTeX ↓Parameter-Efficient Deep Learning Models for Vital Sign Estimation From PPG
Compact deep learning models for estimating heart rate, blood oxygen saturation and respiratory rate from photoplethysmography, with mobile deployment in mind.
Cite this work
@article{samavati2026medvse,
title = {Parameter-Efficient Deep Learning Models for Vital Sign Estimation From PPG},
author = {Taha Samavati and Mahdi Farvardin and Aboozar Ghaffari},
journal = {Healthcare Technology Letters},
year = {2026},
volume = {13},
number = {1},
eid = {e70083},
doi = {10.1049/htl2.70083},
url = {https://doi.org/10.1049/htl2.70083}
}
Sparse 3D Perception for Rose Harvesting Robots: A Two-Stage Approach Bridging Simulation and Real-World Applications
A two-stage vision approach for detecting roses and estimating their 3D location, connecting synthetic training environments with real-world robotic harvesting.
Cite this work
@misc{samavati2025rose3d,
title = {Sparse 3D Perception for Rose Harvesting Robots: A Two-Stage Approach Bridging Simulation and Real-World Applications},
author = {Taha Samavati and Mohsen Soryani and Sina Mansouri},
year = {2025},
eprint = {2508.00900},
archivePrefix = {arXiv},
doi = {10.48550/arXiv.2508.00900},
url = {https://arxiv.org/abs/2508.00900}
}
LDPSA: A Large Dataset of Persian Sentiment Analysis
A large Persian-language dataset supporting sentiment analysis research and the evaluation of natural language processing models.
Cite this work
@inproceedings{samavati2024lpdsa,
title = {LDPSA: A Large Dataset of Persian Sentiment Analysis},
author = {Ali Nazarizadeh and Taha Samavati and Sina Mansouri and Minoo Sayyadpour},
booktitle = {2024 10th International Conference on Web Research (ICWR)},
year = {2024},
pages = {375--381},
doi = {10.1109/ICWR61162.2024.10533380},
url = {https://doi.org/10.1109/ICWR61162.2024.10533380}
}
Deep learning-based 3D reconstruction: a survey
A survey of deep learning approaches to reconstructing 3D shape, organizing methods, representations and the challenges of recovering geometry from observations.
Cite this work
@article{samavati20233drec,
title = {Deep learning-based 3D reconstruction: a survey},
author = {Taha Samavati and Mohsen Soryani},
journal = {Artificial Intelligence Review},
year = {2023},
volume = {56},
number = {9},
pages = {9175--9219},
doi = {10.1007/s10462-023-10399-2},
url = {https://link.springer.com/article/10.1007/s10462-023-10399-2}
}
Text as environment: A deep reinforcement learning text readability assessment model
An approach to assessing text readability that treats text as an environment for a deep reinforcement learning model.
First posted 2019 · Latest revision October 2023
Cite this work
@misc{samavati2019textasenv,
title = {Text as environment: A deep reinforcement learning text readability assessment model},
author = {Hamid Mohammadi and Seyed Hossein Khasteh and Tahereh Firoozi and Taha Samavati},
year = {2019},
eprint = {1912.05957},
archivePrefix = {arXiv},
doi = {10.48550/arXiv.1912.05957},
url = {https://arxiv.org/abs/1912.05957}
}
Publication records checked against publisher and arXiv sources on 2 November 2026. Preprints and later journal versions are grouped as one work. Citation metrics are a dated Scholar snapshot.